The State of Marketing Mix Modeling Deck - Council for Research
Transcript of The State of Marketing Mix Modeling Deck - Council for Research
The Current State Of Marketing Mix Models
Presented by
September 20, 2013
Presented by
September 20, 2013
2
The Council For Research Excellence
• Consists of 35+ senior-level research professionals.
• Represents advertisers, agencies, networks, cable companies, and station groups.
• Seeks to advance the knowledge and practice of methodological research.
3
CRE Member Companies
3
4
Agenda
Welcome and Introduction
Richard Zackon (CRE), Dave Poltrack (CBS, CRE)
Presentation
Jim Spaeth, Alice Sylvester (Sequent Partners)
Update on Industry Initiatives
Gayle Fuguitt (ARF), Bob Liodice (ANA), Mike Donahue (AAAA), George Ivie (MRC), Dave Poltrack (CRE)
Panel
Jim Spaeth (Sequent), Doug Jensen (Avon), Larry Davis-Swing (MediaVest), Dave Poltrack (CBS), Neil Canter (Nielsen)
Q&A
CRE ROI Committee
Members:
• Michele Buslik, Targetcast• Laura Cowan, Lin Media• Ed Gaffney, Group M • Jeffrey Graham, Twitter • Mark Kaline, KCC• Doug Pulick, NCM • Bryon Schafer, Hulu
Chair: Dave Poltrack, CBS Corporation
• Ceril Shagrin, Univision• Howard Shimmel, Turner
Broadcasting• Kate Sirkin, MediaVest• Leslie Wood, Nielsen• Richard Zackon, CRE• Tom Ziangas, AMC
6
Project Objectives
• Provide an in-depth understanding of current methods used to estimate advertising ROI
• Assess the strengths and weakness of the data, techniques, and organizational processes employed
• Begin defining best practices
• Recommend improvement priorities
7
9
40 Interviews
Among Leading Practitioners, Academics and Advertisers
Discussed: • Models In Use Today• Quality Of Media Inputs• Application Of Model Outputs• Beginning Of Best Practice Discussion
Unprecedented Complexity
• Increasingly sophisticated models
• Highly variable processes for applying the learning
• Models evolving in light of complex media world and focus on “funnel metrics”
• Challenging media data acquisition
11
Range of Sophistication
• For modelers who are constrained by time/money demands, straightforward models fit their needs
• Other modelers are pushing boundaries ─ wrestling with long term, complex interactions and more complete view of the contribution of advertising
12
13
Big Split At The Advertisers
Two Distinct Users
• Analytics/modeling people who really get it
• Less knowledgeable users• Not equipped to address data or
technical issues• Learned to trust the modelers over
time • Assume the models are right
Models In Use Today
14
• Regressions
• Series of Equations
• Agent Based Models
• Single Source
• Cross-Channel Attribution
• Highly Efficient; Collinearity Issues
• Incorporate brand effects and interactions
• Reflect complexity of networks and systems
• Uniquely isolate advertising effects
• Digital only; isolated
Observations:Approach:
• Advertising components deployed against different objectives
• Short-term financial performance • Contribution margin and revenue-
based returns are used • Allows modelers to align
recommendations with business strategy
• Some models use consumer metrics like awareness, preference, consideration etc.
15
What’s Modeled?
16
Long Term Effects
• ROO as well as ROI
• Upper funnel effects – awareness consideration
• Longer term Adstocks, repeat purchase patterns
Just Now Beginning To Be Addressed …Good Time For Consensus
• TV copy test data as diagnostics• Cascade of digital behaviors captured
and may act as copy quality measure• Quality of ads in other media
generally not considered
17
Creative Quality Not Uniformly Considered
…Despite the fact that creative quality can explain 70% of advertising’s effects
• Weekly • Campaign or creative execution level • Reach/frequency delivery • By DMA, or even more granular • Consistent for all media elements
18
Media Data Inputs
Best Data Has These Characteristics:
19
Modelers Are Not Media Researchers
• Planned vs. delivered
• Spend vs. GRPs or Impressions
• GRPs vs. TRPs vs. reach & frequency
• Broad averages vs. specific ad exposures
With Some Exceptions Generally Use What They Get … Don’t Know What They Don’t Know:
• TV – GRPs by market and week are generally available; model at the daypart, length and campaign level when possible
• Radio – New PPM TRP data available by week, but not widely recognized or in use
• Print – Improvements made to provide best estimates of GRPs by market and week, recognized but not widely used
20
Media Inputs In Use
Traditional Media
• Mobile – Not currently modeled for the most part; quite a bit of work required
• Social – Many metrics; struggling to find meaning
• Digital – Most difficult; abundant data, attribution concerns; situation may be improved with OCR; area of most development work
21
Media Inputs In Use
New Media
• Hispanic – No unique issues• Other Media –
(Buzz, Events) – elusive, not yet part of mainstream
22
Media Inputs In Use
Other Media
• Media data acquisition process is unduly onerous, sloppy and not standardized
• Represents a huge amount of time and effort
23
Ugly Data Acquisition
24
Best In Class Process
• Cross-functional team with consistent involvement from the marketer, modeler and agencies
• Modeling findings integrated with other sources such as tracking studies and copy tests
• Finance plays a role - applies modeling results to budgeting decisions
• CEO or CMO support ensures modeling findings and recommendations are adopted
To Drive Advertising Decisions
• Big Range• Some simply supply data; later
informed of results• Others plan alternatives based on
model results• Others routinely interact throughout
the process
• Creative agencies have no real seat at the table
• Not sure there is a re-entry process once media don’t perform well in the models• Role for “Test and Learn”
25
Agency Involvement In The Process
• Simplify, standardize media data• Establish best
practices• Encourage their
development• Audit their use
26
Opportunities – Media Data
• Encourage open R&D toward measuring long-term effects of advertising
• Underscore importance of measuring creative efficacy – in all media
27
Opportunities – Techniques
• Begin dialogue around organizational issues with applying model results to advertising and media decisions• Update ARF best
practices• Engage 4As and ANA• Provide user
education28
Opportunities – Organizational Process
31
Industry Response
Mike DonahueEVP, Strategic Partnerships,American Association of Advertising Agencies
35
Panel Discussion
• Jim Spaeth, Sequent Partners, Moderator
• Doug Jensen, Global Marketing Intelligence, Avon
• Larry Davis-Swing, SVP, Director of Analytics, MediaVest
• Dave Poltrack, Chief Research Officer, CBS
• Neil Canter, President, Global Analytic Consulting, Nielsen